8596
Biochemistry 1993, 32, 8596-8601
Photoaffinity Labeling of the Vesamicol Receptor of Cholinergic Synaptic Vesiclest Gary A. Rogers' and Stanley M. Parsons Department of Chemistry and the Neuroscience Research Institute, University of California, Santa Barbara, California 93106 Received March 1, 1993; Revised Manuscript Received June 2, 1993
ABSTRACT: On the basis of the high-affinity vesamicol analog 4-aminobenzovesamicol (ABV), a tritiated, arylazido ligand (azidoABV) of thevesamicol receptor (VR) in cholinergic synaptic vesicles was synthesized. azidoABV is an inhibitor of acetylcholine (AcCh) active transport, and it binds to the VR with higher affinity thanvesamicol. The rate of dissociation of azidoABV from synaptic vesicles is 0.058 f 0.003 min-* at 20 OC (about 3-fold slower than that of vesamicol), and the equilibrium dissociation constant is 2 nM (about 4-fold lower than that of vesamicol). Photolysis of [3H]azidoABV in the presence of a stoichiometric excess of the VR led to incorporation of 28% of the radiolabel, of which 57% was blocked by 50 /IM vesamicol. Sodium dodecyl sulfate polyacrylamide gel electrophoreticanalysis of the labeled vesicles revealed, after autofluorography, specific labeling over a broad molecular weight range that extended from about 50 to 200 kDa. This labeling pattern was essentially the same as that obtained with an azido analog of AcCh that was used to label the AcCh transporter (Rogers, G. A., & Parsons, S . M. (1992) Biochemistry 31, 5770-5777). In addition, about 6% of the radioligand that was specifically incorporated into proteins with M , greater than 12 kDa labeled four polypeptides that corresponded to bands in the Coomassie image at M , = 23, 33, 35, and 38 kDa. The results suggest that the VR exists as part of a complex system of subunits.
We recently reported the results of a successful effort to photoaffinity label the acetylcholine transporter (AcChT)' of cholinergic synapticvesicles using a radiolabeledarylazido derivative of an acetylcholineanalog [azidoAcCh (Rogers & Parsons, 1992)l. That ligand covalently labeled in a pharmacologically relevant manner a vesicular component with diffuse electrophoretic migration very similar to that of the vesicular proteoglycan (VPG). The AcChT is reversibly inhibited by a large family of drugs based on the vesamicol structure [ (-)-trans-2-(4-phenylpiperidino)cyclohexanol (Rogers et al., 1989)] that binds at a site (the vesamicol receptor, VR) allosteric to the AcCh transport site (Bahr et al., 1992a). The VR has been purified in a detergent-solubilized form, and it also exhibits diffuse electrophoretic migration similar to that of the VPG (Bahr & Parsons, 1992). The VR copurifies with two epitopes, namely SV1 and SV2 (Bahr & Parsons, 1992), and exists in a very tight linkage with the VPG as confirmed by immunochemical and enzymatic techniques (Bahr et al., 1992b). The SV1 epitope is luminal and resides on a keratan sulfate-like (Scranton et al., 1993; Bahr et al., 1992b) portion of the VPG. The SV2 epitope is cytoplasmic and is found on secretory vesicles in essentially all neuronal and endocrine cell types (Buckley & Kelly, 1985) and, therefore, is not cholinergic. These are surprising results that prompted us to further secure the identification of the VR by photoaffinity labeling. We report confirming results in this paper, as well as the labeling of other polypeptides that are possibly associated with the VR. 7 Supported by Grant NS15047 from the National Institute of Neurological Disorders and Stroke and a grant from the Muscular Dystrophy Association of the USA. Abbreviations: AcCh, acetylcholine; AcChT, acetylcholine transporter; vesamicol, (-)-trans-2-(4-phenylpiperidino)cyclohexanol; VR, vesamicol receptor; VPG, vesicular proteoglycan; DTT, dithiothreitol; ABV, 4-aminobenzovesamicol; GlyABV, N-glycyl-4-aminobenzovesamicol; azidoABV, N-(azidobenzoylglycyl)-4-aminobenzovesamico~ azidoAcCh, photoaffinity acetylcholine analog; kq, pseudo-first-order rate constant.
MATERIALS AND METHODS General Methods. 'H NMR spectra were recorded on a General Electric GN500 with tetramethylsilane as internal standard. UV spectra were recorded on a Shimadzu UV265 spectrophotometer. Mass spectra were determined on a VG Analytical 70-250 H F mass spectrometer in the FAB or E1 mode. Autofluorographswere scanned and quantitated using an LKB Ultroscan XL. Column chromatography was performed on Merck silica gel 60. All chemicals and reagents were of the highest commercialquality. All calculations were performed using the commercial software MINSQ (MicroMath Scientific Software, Salt Lake City, UT) and the appropriate equations. Derived parameters are quoted f 1 standard deviation. Resting synaptic vesicles (VP,) were isolated from the electric organ of Torpedo calijornica as previously described (Yamagata & Parsons, 1989), with the modification that concentration of the vesicles after size exclusion chromatography on Sephacryl 1000 was carried out using Centriprep-30 centrifugal ultrafiltration concentrators (Amicon Corp.) at 4 OC, unless stated otherwise. After concentration of the vesicles, the pH of the isolation buffer (7.0) was adjusted to 7.8 by addition of the required amount of buffer A at pH 8.2 (adjusted with KOH) that was made up of 100 mM HEPES, 700 mM glycine, 1 mM EDTA, and 1 mM EGTA. All subsequent experiments in buffer A were at pH 7.8. All ligand-binding assays were performed by vacuum-assisted filtration onto prewetted polyethylenimine-treatedglass-fiber filters (Whatman GF/F, 1.3 cm) that were then washed three times with ice-cold buffer unless stated otherwise. Tritium was quantitated by liquid scintillation spectroscopy. Protein concentration was determined according to the method of Bradford (1976) using bovine serum albumin as a standard. N-(4.'-AzidobenzoylglycyI)-4-aminobenzovesamicol (azidoABV). Succinimidyl4-azidobenzoate(Pierce) (95 mg, 365 pmol) was partially dissolved in 2 mL of dry CH2C12. Racemic N-glycyl-4-aminobenzovesamicol[GlyABV (Rogers et al.,
0006-2960/93/0432-8596%04.00/0 0 1993 American Chemical Society
Photoaffinity Labeling of the Vesamicol Receptor
Biochemistry, Vol. 32, No. 33, 1993 8597
A that contained 3 mM MgC12) were incubated with 4.8 nM 1993)] (1 50 mg, 396 pmol) was added to the solution, which [3H]azid~ABV (in glass treated with Sigmacote) at 20 OC for subsequently became homogeneous. The solution was left in 60 min. Dissociation of [3H]azid~ABVwas initiated by the the dark for 24 h, after which time thin-layer chromatography addition of a vesamicol solution so that the final concentrations (silica gel; 3% ethanol/CHCls) indicated that the reaction were 4.62 nM and 30pM, respectively. [3H]azidoABVbound was complete. Solvent was removed on a rotary evaporator, to vesicles was assayed by filtration at various times as and the residue was refluxed with methanol for 15 min in an described above. attempt to destroy any ester formed by the acylation of the alcohol moiety of the vesamicol ring. Notwithstanding this Rate of Association of [3H]azidoABV with Synaptic effort, silica gel chromatography (CHC13) provided two Vesicles. Synapticvesicles (about 0.6 pg protein/mL in buffer distinct fractions, the first of which was diacylated GlyABV Aat 20 "C) weremixed witha smallvolumeof [3H]azidoABV (FABMS base peak at m / z = 670 for P 1). Diacylation (in glass treated with Sigmacote) to yield a final concentration was confirmed by 'H NMR. The second chromatography of 3.7 nM. After 15-20 s of vigorous vortexing, 200-pL fraction (97 mg) was identified as the desired product by aliquots were taken for the determination of bound radioligand analysis: FABMS, base peak at m / z = 525 for P 1 and 497 by the filter assay described above. for loss of N2; UV, loss of absorbance at A, = 268 nm upon Reversible Binding of [ j H ] azidoABVto Synaptic Vesicles. exposure to a lamp that emits light at 254 nm; IR, strong Synaptic vesicles from the fraction represented by the center absorbances at 1688 (coincident with GlyABV) and 1645 of the chromatography peak from the size exclusion chrocm-l for amide I bands and 2108 cm-* for the azido group. matography step of purification (as described above) were Tritiated (-)-N-(4'-Azidobenzoylglycyl)-4-aminobenzoused without Concentration. The pH of the solution was vesamicol ( [3H]azidoABV). N-[ben~oate-3,5-~H2lSuccin- adjusted to 7.8 by addition of buffer A at pH 8.2. Final imidyl4-azidobenzoate (5.09 nmol, 49.10 Ci/mmol, Du Pont vesicle protein concentration was about 30 pg/mL. The VR NEN) was combined with 10.7 pg (28.2 nmol) of (-)-GlyABV concentration was determined by the addition of 60 pL of (Rogers et al., 1993) in 100 pL of 2-propanol and placed in vesicles to 300 pL of [3H]vesamicol (about 120 nM). After the dark. After 4 days the sample was diluted with 100 pL 30 min of incubation, 100-pL aliquots were assayed for bound of ethanol and kept at -80 OC until subsequent workup. tritium and 10-pL aliquots for the total [3H]vesamicol in Analysis of the crude reaction mixture by C18 reversedsolution. The same procedure was performed with the addition phase HPLC (20% 25 mM, pH 3.2 phosphate buffer in of [1H]vesamicol to a final concentration of 28 pM in order acetonitrile) demonstrated that 93% of the radioactivity eluted to assess nonspecific binding. coincident with an authentic sample of nonlabeled azidoABV. On the basis of the concentration of the VR determined Purification of the product on a semiprep Chiralpak AD above, vesicles were diluted about 60-fold for the binding (Daicel Chemical Ind.) column (25% ethanol/hexane at 4 experiments with [3H]azidoABV. All solutions of [3H]mL/min) afforded 160 pCi (64% radiochemical yield) of azidoABV were made in glass that had been treated with material that eluted in the region of the chromatogram Sigmacote as described above. Duplicate samples of vesicles coincident with theretentiontimeof (-)-azidoABV (26.3 min). were diluted with one-tenth volumes of various stock conThe exact elution time of the product could not be recorded centrations of [3H]azidoABV to give final concentrations of because the UV detector was turned off as soon as elution of ligand that ranged from 59 pM to 8 nM. After an ll-h the photosensitive azidoABV began. Unreacted (-)-GlyABV incubation period, bound tritium was determined as above. was observed to elute at 15.7 min. Actual final total concentrations of radioligand in each sample Inhibition of AcCh Active Transport by azidoABV. Active were determined by liquid scintillation spectroscopy at the transport of AcCh was constituted as described (Rogers et completion of the incubation period. In samples used to al., 1993) at a synaptic vesicle concentration of 0.20 mg of measure nonspecific binding of [3H]azidoABV, (-)-4-amiprotein/mL in buffer A. A 30 mM stock solution of azidoABV nobenzovesamicol (ABV) was added to yield a final concenwas made by dissolving 5.6 mg in 50% DMSO/O.l 1 M HCl. tration in 100-fold excess of the radioligand. Serial dilutions were made into 11% DMSO/H20 and Photoaffinity Labeling of Synaptic Vesicles with [3H]subsequently into 0.4 M KC1 so that the final concentrations azidoABV. Volumes of 156,31, and 6.25 pL of an ethanolic in the transport assay solutions were 0.30 nM to 30 pM solution of [3H]azidoABV were evaporated under a stream azidoABV and a maximum of 0.16% DMSO. of N2 in three glass test tubes that had been treated with Inhibition of [ 3 H l Vesamicol Binding by azidoABV. SynSigmacote. To the first test tube was added 2.4 mL of synaptic aptic vesicles (0.20 mg of protein/mL in buffer A) were vesicles that were 0.65 mg of protein/mL in buffer A. To the incubated with differing concentrations of azidoABV (0.30 second was added 8.33 pL of a 3 mM solution of vesamicol nM to 30pM) and2nM [3H]vesamicol. Bound [3H]vesamicol in buffer A followed by 500 pL of the same vesicle suspension. was assayed by vacuum-assisted filtration through PEI-treated The final concentration of vesamicol was 50 pM. To the glass-fiber filters as described (Rogers & Parsons, 1992). third tube was added 6.25 pL of a 27 pM ethanolic solution Nonspecific binding of [3H]vesamicol to vesicles was deterof racemic ABV, which was also evaporated before the addition mined in the presence of a 100-fold excess of [lH]vesamicol. of 100 pL of vesicle suspension. The final concentration of Binding of [3flazidoABV to Container Surfaces. A 50 ABV was 1 pM. A fourth tube was constituted by incubating nM solution of [3H]azidoABV in buffer A was placed in a 100 p L of the vesicle suspension with 1 pM ABV for 1 h prior polypropylenetube. The tritium concentration was monitored to introduction of the [3H]azidoABV as above. All four as a function of time and vortexing. In a similar manner, a solutions were incubated for 10 min and then illuminated 100 nM solution was placed in standard borosilicate glass with an 18-W UV lamp (Ultra-Violet Products Inc., San tubes that were either untreated or treated with Sigmacote Gabriel, CA) that emits light with intensity centered at 254 (Sigma Chemical Co.) and assayed as a function of time. nm. For 10 min the tubes were alternatively illuminated and Sigmacote produces a coating of siliconeon container surfaces. vortexed to yield a total illumination time of about 5 min each. Following the photolysis, 5 pL of the solution was Rate of Dissociation of [3H]azidoABVfrom Synaptic removed from each tube and diluted into 145 pL of buffer A Vesicles. Synaptic vesicles (20 pg of protein/mL in buffer
+
+
Rogers and Parsons
8598 Biochemistry, Vol. 32, No. 33, 1993 containing 60 pM vesamicol. From each of the diluted samples, 100 pL was vacuum-filtered through PEI-treated glass-fiber filters and washed with 1 mL of buffer A. Each filter was incubated two times with 1-mL volumes of buffer A (containing 60 pM vesamicol) for 10 min at 23 OC. Finally, each was washed with three 1-mL volumes of buffer A, and the bound tritium was determined as before. SDS-PAGE and Autofluorography of Photolabeled Synaptic Vesicles. Volumes equivalent to 30 pg of protein were removed from each sample of photolabeled vesicles from above and diluted to 150 pL with buffer A. Remaining vesicles were set aside and frozen for potential later analysis. The diluted vesicles were pelleted in an Airfuge (Beckman Instrument Co.) at 73 000 rpm for 70 min. The supernate was carefully removed, and the pellets were covered with 20 pL of treatment buffer that contained 10% sodium dodecyl sulfate (SDS) and 20% glycerol in 0.125 M Tris-HC1 (pH 6.8). The samples werevortexed occasionally over a 6-h period at 23 OC. Finally, each was diluted with 20 pL of a 10% (w/v) solution of dithiothreitol (DTT) and heated in a 90 OC bath for 3 min. Molecular weight standards were treated with the same solutions but were given only a brief incubation prior to addition of the DTT. The dissociated samples were subjected to SDS-PAGE using a 5-1 5% linear gradient resolving gel, a 3% stacking gel, and the discontinuous buffer system of Laemmli (1970). Electrophoresis was halted when the tracking dye was about 1 cm from the bottom of the gel. After electrophoresis, the gel was stained with Coomassie Brilliant Blue R in order to visualize proteins and then treated with Intensify (Du Pont) before being dried onto filter paper under vacuum. The dried gel was autofluorographed with XAR-5 film (Kodak) at -80 OC for 5 days.
RESULTS Synthesis and Photolability of azidoABV. From previous studies (Rogers et al., 1993) it was apparent that 4-aminobenzovesamicol (ABV) with a dissociation constant of 6.5 pM might provide the basis for the synthesis of a successful photoaffinity reagent for the VR. Toward this goal, the chemical reactivity of ABV was increased by the addition of a glycyl moiety [GlyABV (Rogers et al., 1993)] to yield an aliphaticprimary amine of about 5 orders of magnitude greater basicity than the anilino nitrogen. However, the secondary alcohol of ABV, which is an essential pharmacophore, has much greater reactivity in acylation reactions than expected, perhaps due to the proximity of the tertiary amine. The hyperreactivity caused facile diacylation of GlyABV, and because of this the yield of racemic azidoABV was only 51%. GlyABV was resolved by chiral chromatography and acylated by tritiated succinimidyl 4-azidobenzoate, which provided enantiomerically pure (-)-[3H]a~id~ABV in 64% radiochemical yield of 49.1 Ci/mmol specific activity. Because the absorption maximum of azidoABV occurs at 268 nm, it was necessary to use a short-wavelength lamp that emits light with intensity centered at 254 nm (18 W). azidoABV in aqueous solution decomposed with a half-life of about 1 min in the configuration utilized here. It is stable under ordinary fluorescent illumination and therefore can be used conveniently in reversible binding experiments. Inhibition of [3HjAcChActive Transport and ['HI Vesamicol Binding by azidoABV. As an analog of vesamicol, azidoABV should be a potent inhibitor of AcCh active transport in synaptic vesicles. When tested for transport inhibition under conditions where the VR concentration was about 60 nM, (f)-azidoABV inhibited active transport with
100
3
60
K
20
a
0
Conc. of AzidoABV (M) FIGURE1: Inhibition of AcCh active transport and [3HH]vesamicol binding by azidoABV. AcCh active transport was measured at 22 "C as previously described (Rogers et al., 1989). Synaptic vesicle protein and ["H]AcCh concentrations were 0.20 mg/mL and 50 pM, respectively. The best hyperbolic fit to the data (- 0 -) using nonlinear regression analysis gave an ICs0value of 150 60 nM. Inhibition of [3H]vesamicolbinding to vesicles was measured at 22 OC as described in Materials and Methods. VR and [3HH]vesamicol concentrationswere 67 and 2 nM, respectively. The best hyperbolic fit to the data (- - - -) gave an ICs0value of 600 200 nM. About 98% of the specifically bound [3H]vesamicolwas displaced.
*
*
an apparent ICSovalue of 150 f 60 nM (Figure 1). In the same preparation of vesicles, azidoAcCh exhibited an ICs0 value more than 5 times greater (data not shown). In competition with 2 nM [3H]vesamicol bound to synaptic vesicles with a VR concentration of 67 nM, (A)-azidoABV exhibited an apparent ICs0 value of 600 f 200 nM and displaced 98% of the specifically bound vesamicol (Figure 1). The ICs0values for the inhibition of AcCh active transport and vesamicol binding are denoted as apparent because subsequent experiments with radiolabeled azidoABV demonstrated that there is a time-dependent loss of azidoABV to glass surfaces, as discussed below. Therefore, the concentrations of unlabeled ligand were significantly lower than anticipated, which produced artifactually high values of IC50. However, the data demonstrate that azidoABV is a potent inhibitor of AcCh active transport, as expected for a vesamicol analog, and it is significantly more potent than azidoAcCh. Binding of azidoABV to Containers. Initial kinetic experiments on the binding of [3H]azidoABV to vesicles suggested complex behavior. Analysis of the solubility properties of the ligand led to the discovery that [3H]azidoABV deposits on glass surfaces in a time-dependent manner that is likely dependent on the surface-to-volume ratio. In an untreated borosilicate test tube, 67% of the [3H]azidoABV in 1 mL of a 100 nM solution deposited on the glass with a half-life of about 35 min (data not shown). Polyethylene also adsorbs the ligand, but mixing appeared to cause partial redistribution from the plastic surface back into solution. Because the affinity for glass could be blocked substantially by treatment with Sigmacote, all reported experiments involving [3H]azidoABV were conducted in treated glass. Binding Kinetics of [3HjazidoABV. The rate constant for dissociation of bound [3H]azidoABV from synaptic vesicles was determined by displacement with 30 pM vesamicol (Figure 2A). At 20 "C the process was well-described by a single exponential decay with k-1 = 0.058 f 0.003 min-1 (t1p = 12 min). The rate of association of [3H]azidoABVwith vesicles was determined at 20 "C under pseudo-first-order conditions (i.e., the concentration of free ligand remained essentially constant during the process) (Figure 2B). Under these conditions, k*
Photoaffinity Labeling of the Vesamicol Receptor
Biochemistry, Vol. 32, No. 33, 1993 8599 A
zg
O0.0
10.0
m
20.0
90.0
z
40.0
Time (min)
0.400
0.000
T
0.0
I
2.0
Conc. of
4.0
8.0
8.0
10.
[%] AzidoABV (nM)
FIGURE 3: Reversible binding of [)H]azidoABV to synapticvesicles. Vesicles were constituted as described in Materials and Methods. Bound [3H]azidoABVwas determined in the presence (- - - -) or absence (- 0 - ) of a 100-fold excess of ABV to yield curves representing the amount of nonspecificand total bound, respectively. A simultaneous fit of a linear equation (for nonspecific binding) and an equation combining the linear equation with a hyperbolic term (for total binding) to the separatedata sets yielded values of KO and B , of 2.1 f 0.4 and 0.29 0.03 nM (about 580 60 pmol/mg of protein), respectively.
*
Table I: Analysis of Photolabeling by Filter Assay
20 4 0.0
5.0
10.0
15.0
I
20.0
Time (min) FIGURE 2: Kinetics of interaction of ['HIazidoABV with synaptic vesicles. (A) Dissociation of 4.6 nM ['HIazidoABV from vesicles (20 pg of protein/mL) was promoted by 30 pM vesamicol at 20 OC. About 92% of the total bound tritium dissociated by a process that is described by a single exponential equation with k = 0.058 f 0.003 m i d . Similar results were obtained with two other preparationsof vesicles. (B) Association of 3.7 nM ['HIazidoABV with synaptic vesicles (about 0.6 pgof protein/mL) at 20 "C.The VRconcentration was about 0.18 nM as measured by [3H]vesamicolbinding, and at equilibrium only 0.1 nM drug was bound to the VR. The association process was well-described by an equation with a single exponential and a pseudo-first-order rate constant (k9) = 0.14 0.03 min-I.
*
*
= 0.14 0.03 m i d , and as the ligand concentration was 3.7 nM, the second-order rate constant (k+/[azidoABV]) = 3.8 X 107 M-l min-1. Therefore, the value of KD determined from the rate processes (k-l/(k+/[azidoABV])) = 1.5 nM. azidoABV to Synaptic VesEquilibrium Binding of icles. The reversible binding of [3H]azidoABV to vesicles was assessed after an 11-h incubation period at 23 O C (Figure 3). Based on the concentrations employed and the rate constants determined above, the approach to equilibrium should have been greater than 90% completed at the lowest concentrations of the ligand that were used. Under these conditions, values for the parameters KD and B,,, were determined to be 2.1 f 0.4 and 0.29 f 0.03 nM (or about 580 f 60 pmol/mg at about 0.5 pg of protein/mL), respectively. Photolabeling of Synaptic Vesicles with [3flazidoABV. The synaptic vesicles that were used for the photolabeling experiment were assayed by [3H]vesamicol (250 nM) and found to possess 690 pmol of VR/mg of protein. At the protein concentration of 0.65 mg/mL, therefore, the VR concentration was 450 nM. The measured concentrations of [3H]azidoABV in solution in the four samples are listed in Table I. The amount of photolabeling was analyzed by two methods. The first method was by filter assay followinga 30-fold dilution
1 none 128 65 28 2 50 pM vesamicol 154 31 12 3 lpM(&)-ABV 151 71 26 4 1 pM (&)-ABV' 136 54 23 * The same vesicles were also analyzed by SDS-PAGE and autofluorography,and the results are shown in Figure 4. b Concentrationsvaried due to adsorption to the glass test tube. Tritium bound to vesicles that could not be removed by exhaustive washing. d Bound ['HIazidoABV divided by the total amount of [3H]azidoABVthat was present. Added 1 h prior to the addition of [3H]azidoABV.
into a 60 pM vesamicol solution. In addition, two 10-min incubations on the filters with 60 pM vesamicol were used in order to promote dissociation of specifically bound but not covalently incorporated ligand. Kinetic data presented in Figure 2A demonstrate that the high concentration of vesamicol should have promoted nearly complete dissociation of reversibly bound [3H]azidoABVfrom the VR. The amounts of tritium that remained bound to the filters and were presumably covalently incorporated into the vesicles are listed in Table I. The values for the percent incorporation are also listed. These data suggest that over 50% of the total labeling was blocked by a large excess of vesamicol but that much less was blocked by a small excess of the much more potent ligand ABV. The second method of analysis was by gradient SDS-PAGE (5-1596) followed by staining with Coomasie Blue and autofluorography (Figure 4). Protein staining revealed wellfocused bands throughout the molecular weight range of 12200 kDa. Autofluorography of the unprotected sample (sample 1) demonstrated heavy labeling of a continuous region from about 50 to 200 kDa. Quantitation of the absorbance indicated that 94% of the total labeling of proteins (exclusive of the labeling at the dye front) was in this region. In addition, four polypeptides with Mi's of about 22.6, 33,35.1, and 37.5 kDa were specifically labeled. The 33-kDa band was somewhat diffuse, but correlation with the Coomassie image suggests that the diffuseness arises as the result of a doublet
8600 Biochemistry, Vol. 32, No. 33, 1993
200 116-
97 77 66 -
55.542.7
-
30 -
-
17.2 Dye
Ql -aP -QlP .-ca, Q. t
2
a
E
0
tn
E
CI
tn
E
0
tn
FIGURE4: Autofluorographic analysis of synaptic vesicles photolabeled with [3H]azidoABV. The amount of tritium covalently incorporatedinto vesicles was also analyzed by the filter assay, and the results are presented in Table I. Photolabeling in the presence or absence of competing ligands and the preparation of vesicles for SDS-PAGE and autofluorographywere performed as describedunder Materialsand Methods. The figure is a composite showingCoomassie Blue staining and a 3-day film exposure at -80 "C. The first lane, labeled Protein, shows the Coomassie staining pattern, which was identical for all four samples. Sample 1 shows the autofluorographic pattern for nonprotected vesicles. Samples 2 and 3 show labeling when 50 pM vesamicol or 1 pM (f)-ABV was added simultaneously with theaddition of [3H]azidoABV,respectively,and Sample 4 shows labeling when 1 pM (h)-ABV was added 1 h prior to the addition of [3H]azidoABV. The M,'s in kilodaltonsand positionsof standard proteins are indicated to the left. Labeled bands at 22.6, 33, 35.1, and 37.5 kDa in Sample 1 were visible in the original photograph but may not be in the published reproduction. of closely-spaced bands at 32.4 and 33.8 kDa. Addition of 50 pM vesamicol coincident with [3H]azidoABVcompletely blocked incorporation of the photoaffinity ligand into components with Mrgreater than 12kDa and somewhatdiminished the labeling of components at the bottom of the gel at the dye front (sample 2). At the lower concentration of 1 p M , (&)ABV protected nearly all sites above 12kDa from being labeled but actually increased incorporation of label into the small components near the dye front, whether it was applied coincident (sample 3) or prior (sample 4) to exposure of the vesicles to the radioligand. Neither vesamicol nor ABV produced any visible change in the protein staining pattern (data not shown).
DISCUSSION Structure-activity studies demonstrated that the benzovesamicol series of analogs has much higher affinity for the VR than does vesamicol (Rogers et al., 1993). Furthermore, earlier studies showed that attachment of a large polar moiety such as biotin to the 4-position of ABV via a tether does not greatly diminish the potency of the analog compared to vesamicol (Rogers et al., 1989). It was reasonable to expect, therefore, than an arylazidomoiety attached to GlyABV would
Rogers and Parsons constitute a high-affinity photolabile probe (azidoABV) suitable for specifically labeling the VR. azidoABV was synthesized as reported above and shown to possess inhibitory properties consistent with those of a vesamicol analog, as both AcCh active transport and [3H]vesamicol binding were inhibited. One of the most useful parameters that characterizes ligand receptor interactions is the half-life for dissociation of the complex. This we determined for the [3H]azidoABV-VR complex to be 12 min at 20 OC, or about 3 times longer than for vesamicol (Rogers et al., 1993). Therefore, azidoABV can be used for reversible binding experiments. The association rate constant of about 4 X lo7M-I min-I is approximately 5-and 2.5-fold lower than those ofvesamicol and ABV, respectively(Rogers et al., 1993). The dissociation constant KD that can be calculated as the ratio of the dissociation and association rate constants is 1.5 nM. The value for K D determined from the equilibrium binding curve is in good agreement at about 2 nM. These data provide evidencethat [3H]azidoABVhas highly desirable kinetic and equilibrium properties for a photoaffinity ligand. The relatively slow dissociation of the ligand from the binding site should allow ample time for the photogenerated nitrene to insert into nearby receptor chemical bonds. A simple and rapid analysisof the photolabeling experiment by filter assay suggested that vesamicol but not ABV blocked labeling of the VR by [3H]azidoABV. That this is not the case can be seen clearly in the analysis of the same vesicles by SDS-PAGE followed by autofluorography. Photolysis of the unprotected sample of vesicles led to intensive labeling of a species that extends from about 50 to 200 kDa. This result is essentially the same as that obtained by labeling vesicles with [3H]azidoAcCh. Thus we conclude that this diffuse, heavily-labeled species is the VR. At first glance the differences in labeling patterns among the protected samples might seem difficult to reconcile with each other and with the filter analysis of the same samples. However, it should noted that the conditions of the experiment (receptor concentration about 3-fold greater than that of [3H]azidoABV)were devised to minimize nonspecific labeling in the unprotected sample and make maximum use of the radioligand. Thus, at a [3H]azidoABV concentration of about 130-1 50nM in the presence of 450 nM VR, nearly all of the ligand will bind to the receptor and very little will remain free in solution. A 300-fold excess of vesamicol over the [3H]azidoABVconcentration, added at the same time as [3H]azidoABV,would be expected to abolish all labeling of the VR, as was observed. However, a less than 7-fold excess of the high-affinity analog ABV ( K D= 6.5 pM), added to the vesicles at the same time as [3H]azidoABV,will allow some specific labeling to occur because the ligands have similar associationrate constants and photolysis was performed before equilibrationof theligands with the VR was established. In contrast, ABV that is added before [3H]azidoABV will occupy all of the VR and more completely abolish labeling, as was observed. It is noteworthy that the concentration of the more potent enantiomer, (-)-ABV, that was used to block specific labeling was 500 nM, or only in slight excess over the concentration of the VR, and so very little (-)-ABV remained free in solution. The next point that deservescomment concerns the intensive labeling of vesicular components near the bottom of the gel. As virtually all of the [3H]azidoABV was bound to the VR in the unprotected sample, the photogenerated nitrene must have labeled lipid or other small componentsin the surrounding membrane even while bound to the VR. More labeling of the small componentswas produced by the presence of a relatively
Photoaffinity Labeling of the Vesamicol Receptor low concentration of ABV. This demonstrates that the increased free (or solution) concentration of [3H]azidoABV, which results as a consequence of displacement from the VR, produces apparently nonspecific labeling of similar small components. Labeling of this material also occurred in the presence of a large excess of vesamicol(300-fold higher concentration than [3H]azidoABV). Even though the free or solution concentrations of [3H]azidoABV in the ABV- and vesamicol-protected samples were equal, the decreased nonspecific labeling of small components in the presence of vesamicol could result from electrostatic screening of the membrane by the relatively high concentration of vesamicol. Vesamicol, ABV, and azidoABV are all positively charged at the pH of the photolabeling experiments. If the labeled low Mr components are lipids, the intensive labeling of these components in the unprotected sample indicates that specifically bound [3H]azidoABVis in contact with the membrane bilayer. About 6% of the specific labeling above 12 kDa was on four or five peptides of about 23,33 (possibly a doublet), 35, and 38 kDa. At least three mechanisms for producing this labeling can be envisioned. By the first, the specifically labeled peptides might arise from proteolytic breakdown of the VR. Although vesamicol binding is extremely resistant to proteolysis (Kornreich & Parsons, 1988), it is possible that proteolysis could produce smaller fragments still able to bind vesamicol. However, in three photolabeling experiments with azidoAcCh, there was no evidence for the presence of proteolytic fragments of the AcChT-VR (Rogers & Parsons, 1992). By the second mechanism, it is conceivable that [3H]azidoABV bound to the VR was oriented so that the nitrene could covalently react with other peptides that the VR randomly collided with as a result of lateral diffusion in the membrane bilayer. If this mechanism obtains, the result suggests that only a few of the many vesicular proteins randomly collided with the VR during the several minutes required for photolabeling. By the third mechanism, the specifically labeled peptides of 23-36 kDa might represent proteins that are associated with the AcChT-VR. As it was present on a flexible tether extending from the tightly bound ABV core, the photogenerated nitrene could range over a
Biochemistry, Vol. 32, No. 33, 1993 8601 region of space adjacent to the vesamicol binding site. This could lead to specific labeling of accessory proteins if they are present in a heterooligomeric complex with the VR. This interpretation is consistent with localization of the vesamicol binding site on the surface of the VR near the bilayer. In contrast, the AcCh binding site has been suggested to be inside of the AcChT in a channel-like structure (Bahr et al., 1992b) where the nitrene derived from azidoAcCh would not be expected to be able to reach accessory proteins. In summary, it is clear that azidoAcCh and azidoABV specifically label the same vesicular component, which exhibits extremely heterogeneous electrophoreticbehavior. The vesa m i d analog yields evidence of interaction with other protein subunits, which suggests that the AcChT-VR is a complex system.
ACKNOWLEDGMENT We thank Daniel Oros and Wendy Connelly for the preparation of synaptic vesicles. REFERENCES Bahr, B. A., & Parsons, S. M. (1992) Biochemistry 31, 57635769. Bahr, B. A., Clarkson, E. D., Rogers, G. A., Noremberg, K., & Parsons, S. M. (1992a) Biochemistry 31, 5752-5762. Bahr, B. A., Noremberg, K., Rogers, G. A., Hicks, B. W., & Parsons, S . M. (1992b) Biochemistry 31, 5778-5784. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254. Buckley, K., & Kelly, R. B. (1985) J . CellBiol. 100,1284-1294. Kornreich, W. D., & Parsons, S. M. (1988) Biochemistry 27, 5262-5267. Laemmli, U. K. (1970) Nature 227, 680-685. Rogers, G . A., & Parsons, S. M. (1992) Biochemistry 31,57705777. Rogers, G. A., Parsons, S. M., Anderson, D. C., Nilsson, L. M., Bahr, B. A., Kornreich, W. D., Kaufman, R., Jacobs, R. S., & Kirtman, B. (1989) J . Med. Chem. 32, 1217-1230. Rogers, G. A,, Kornreich, W. D., Hand, K., & Parsons, S.M. (1993) Mol. Pharm. (in press). Scranton, T. W., Iwata, M., & Carlson, S. S. (1993) J . Neurochem. 61, 29-44. Yamagata, S . K., & Parsons, S. M. (1989) J . Neurochem. 53, 1354-1362.
